OEDIPE, a software for personalized Monte Carlo dosimetry and treatment planning optimization in nuclear medicine:

c EDP Sciences2014

DOI:10.1051/radiopro/2014021

Available online at:

www.radioprotection.org

A rticle

OEDIPE, a software for personalized Monte Carlo dosimetry and treatment planning optimization in nuclear medicine:

absorbed dose and biologically effective dose considerations

A. Petitguillaume

, M. Bernardini

, D. Broggio

, C. de Labriolle Vaylet

, D. Franck

and A. Desbrée

1 IRSN, Laboratoire d’Evaluation de la Dose Interne, 92262 Fontenay-aux-Roses, France.

2 Hôpital Européen Georges Pompidou, Service de médecine nucléaire, 75015 Paris, France.

3 UPMC, Univ. Paris 06 Bio physics, 75005 Paris, France.

4 Hôpital Trousseau, Service de médecine nucléaire, 75012 Paris, France.

Received 4 April 2014 – Accepted 30 July 2014

Abstract – For targeted radionuclide therapies, treatment planning usually consists of the administration of standard activities without accounting for the patient-specific activity distribution, pharmacokinetics and dosimetry to organs at risk. The OEDIPE software is a user-friendly interface which has an automation level suitable for performing person- alized Monte Carlo 3D dosimetry for diagnostic and therapeutic radionuclide administrations. Mean absorbed doses to regions of interest (ROIs), isodose curves superimposed on a personalized anatomical model of the patient and dose- volume histograms can be extracted from the absorbed dose 3D distribution. Moreover, to account for the differences in radiosensitivity between tumoral and healthy tissues, additional functionalities have been implemented to calculate the 3D distribution of the biologically effective dose (BED), mean BEDs to ROIs, isoBED curves and BED-volume his- tograms along with the Equivalent Uniform Biologically Effective Dose (EUD) to ROIs. Finally, optimization tools are available for treatment planning optimization using either the absorbed dose or BED distributions. These tools enable one to calculate the maximal injectable activity which meets tolerance criteria to organs at risk for a chosen fractiona- tion protocol. This paper describes the functionalities available in the latest version of the OEDIPE software to perform personalized Monte Carlo dosimetry and treatment planning optimization in targeted radionuclide therapies.

Keywords: OEDIPE/nuclear medicine/dosimetry/radiobiology/treatment planning

1 Introduction

For decades, targeted radionuclide therapies mostly con- sisted of the use of¹³¹I for thyroid pathologies. Recently, new radiopharmaceuticals have been introduced in clinics to treat different types of cancer (EANM,2013). Their design relies on specific vectors, such as peptide receptors, hormones or an- tibodies, and radionuclides, selected depending on their linear energy transfer, relative biological effectiveness and chemical properties. Associated clinical trials are designed on the same model as those conducted for chemotherapy (Glattinget al., 2013). Standard amounts of activities are administered and increased progressively to test for possible limiting toxicity.

The recommended administered activity is derived from these mean outcomes and eventually adjusted depending on patient- specific parameters, such as the patient’s weight.

Because of the substantial uptake of ¹³¹I by the thy- roid gland, this methodology has led to the high efficacy of treatments for thyroid pathologies with low or no toxicity.

[email protected]

However, the targeting of tumor cells by new radiopharmaceu- ticals is much more complex, and their distribution and bioki- netics are potentially highly patient-specific (EANM, 2013).

Because biological effects, both in terms of response and tox- icity, are primarily dependent on absorbed doses delivered to tissues rather than on the administered activity (EANM,2013), highly personalized techniques estimating absorbed doses de- livered to healthy tissues are crucial. These estimations solely can ensure that healthy tissue irradiation will not lead to unac- ceptable toxicity, and enable treatment planning optimization by the calculation of the maximal activity that could be ad- ministered to each specific patient. This concept was specifi- cally highlighted in the 2013/59/EURATOM directive from the Council of the European Union (CEU,2013). Furthermore, be- yond treatment planning optimization, an accurate knowledge of absorbed doses delivered to targeted and non-targeted tis- sues is essential to establish dose-effect relationships in terms of efficacy and toxicity.

Over the years, different methods have been developed for dosimetry in nuclear medicine. According to the MIRD for- malism (Bolch et al.,2009), absorbed doses are determined

276 A. Petitguillaumeet al.: Radioprotection 49(4), 275-281 (2014) from the cumulated activity and S-factors, which are depen-

dent on the geometry (standard or patient-specific) and radio- tracer properties (radionuclide, physical and biological half- lives, residence times). Basic methods consist of the use of standard S-factors considering homogeneous activity distribu- tions in each region of interest (ROI). These methods are thus limited as they do not account for the patient-specific anatomy, tumoral lesions specificities and activity distribution hetero- geneity.

Nowadays, thanks to 3D medical imaging advances and in- creasing computational power, CT or MRI data can be used to model the patient’s anatomy using patient-specific voxel phantoms. Moreover, the patient-specific activity distribution can be described from single-photon emission computed to- mography (SPECT) or positron emission tomography (PET) data which enable one to account for its heterogeneity. Fi- nally, the pharmacokinetics can be accounted for using either a kinetic model with parameters obtained from blood, urine or feces sampling or a biokinetic model established from se- rial emission data. The introduction of more accurate methods for dosimetry, such as dose kernel methods or direct Monte Carlo calculations (Bolchet al.,1999; Sgouroset al.,2004;

Petitguillaumeet al.,2014), has followed from these techno- logical advances. Additionally, to improve tumor control while preserving healthy tissues, the notion of Biological Effective Dose (BED) (Dale,1985) was introduced to establish fraction- ation strategies in external beam radiation therapies (EBRT).

These radiobiological considerations, which enable one to take advantage of the differences in radiosensitivities and repair time constants between tumoral and healthy tissues, have been shown to be of interest to increase treatment efficacy while keeping a constant incidence and severity of toxicity (RCR, 2006). Recently, growing interest has been shown in radio- biological aspects involved in targeted radionuclide therapies (Sgouroset al.,2004; Cremonesiet al.,2008) because of their potential added value in improving the risk-to-benefit balance.

Furthermore, the notion of Equivalent Uniform Biologically Effective Dose (EUD) (O’Donoghue,1999) was introduced as a useful quantity to compare heterogeneous absorbed dose dis- tributions with a homogeneous distribution that would have the same biological effects. The EUD can thus be used to assess the levels of potential toxicity and tumor control related to a given absorbed dose distribution (Wuet al.,2002) using, for example, the normal tissue complication probability (NTCP) and tumor control probability (TCP) models.

However, up to now, all these advanced techniques and concepts have been, in the majority, only used for research- oriented projects. The development of the OEDIPE software, a French acronym for “tool for personalized internal dose as- sessment”, was carried out to both pursue research projects and provide a user-friendly dosimetry and treatment planning tool for nuclear medicine clinical applications. OEDIPE de- velopment was initiated at the French Institute for Radiation Protection and Nuclear Safety (IRSN) for in vivo measure- ment calibration and internal contamination dosimetry using direct Monte Carlo calculations (Francket al.,2001; de Carlan et al., 2003). It was then continued to perform dosimetry for nuclear medicine applications from patient-specific vox- elized geometry and estimated cumulated activity distribution

(Chiavassaet al.,2006). Recently, OEDIPE was further devel- oped to automatically derive the 3D activity distribution from emission images registered on the patient-specific geometry.

Additional tools were also implemented to provide the max- imal injectable activity (MIA) that could be administered to the patient according to tolerance criteria for organs at risk (OARs), either for mean absorbed doses, dose-volume frac- tions or maximal absorbed doses (Petitguillaumeet al.,2014).

Finally, an additional module was developed to calculate the 3D distribution of the BED and EUD to ROIs. A specific op- timization tool was also implemented to calculate the MIA for fractionated protocols. The aim of this article is to present the functionalities which are available in this latest version of OEDIPE.

2 Materials and methods

OEDIPE is a user-friendly graphical user interface (GUI), developed in Interactive Data Language (ITT Visual Informa- tion Solutions, Boulder, Colorado). Figure1describes the gen- eral principle of the 3D personalized Monte Carlo dosimetry method. First, ROI outlines, drawn on the patient anatomical images, and registered emission (SPECT or PET) data are di- rectly imported into OEDIPE to define the patient’s anatomy and cumulated activity distribution, respectively. These data are then used in OEDIPE, along with radionuclide properties, to generate the input file for the MCNPX Monte Carlo trans- port code. OEDIPE is finally used to process the MCNPX out- put file. Specific features of the OEDIPE software are further described in the following paragraphs of this section.

2.1 Geometry definition

Patient-specific voxel phantoms can be created from the patient CT or MRI data using OEDIPE. To create these phan- toms, the first method consists of direct segmentation based on thresholds defined in Hounsfield unit values. Otherwise, ROI outlines, generated using external software, can be imported in RTSTRUCT format (Isogray^r, Hermes^r, Pinnacle^r, Inte- grated Registration^r, etc.), .aql format (Aquilab^r) or Dosigray format (Dosigray^r). Once generated, the voxel phantom is dis- played on the phantom viewer interface, presented in Figure2.

This phantom can then be modified using several tools, such as air removal, resizing or fusion of ROIs. Finally, specific mate- rials, defined in terms of density and elemental compositions, can be attributed to each ROI from a database which can be modified by the user.

2.2 Radioactive source definition

Nuclear decay data are directly available in OEDIPE us- ing a database containing 246 radionuclides described in ICRP publication 38 (ICRP, 1993). The radioactive source is then defined either as homogeneous sources located in specific ROIs or as a heterogeneous 3D distribution. For homogeneous sources, the definition of a source only requires one to specify the ROI where it is located, the radionuclide and its cumu- lated activity, expressed in terms of cumulated disintegrations

Fig. 1.General principle of 3D personalized Monte Carlo dosimetry using OEDIPE software.

Fig. 2.Phantom viewer interface with control tools for the display of the voxel phantom.

278 A. Petitguillaumeet al.: Radioprotection 49(4), 275-281 (2014)

Fig. 3.3D viewer interface for the display of isodose or isoBED curves superimposed on the voxel phantom.

occurring in the ROI over a given time period. A heteroge- neous 3D distribution is defined as a 3D matrix of cumulated activity. Counts in each voxel can be obtained from emission data (SPECT or PET) using two methods. First, emission data can be registered beforehand, with external software, on the anatomical images used to create the patient-specific voxel phantom and then uploaded into OEDIPE in DICOM format.

Second, emission data can be registered using OEDIPE’s reg- istration module, which has been developed for rigid registra- tion of SPECT/CT or PET/CT data on a CT or MRI scan. The radionuclide and either the activity in the field of view or a conversion factor must then be keyed in to convert counts into cumulated activity over an infinite period of time expressed in terms of cumulated disintegrations occurring in each voxel.

Finally, cumulated activities in specific ROIs can be removed depending on the user’s intent.

2.3 MCNPX input file generation

After the definition of the geometry and the source dis- tribution, the last parameters that have to be defined are the types of results which are expected from the MCNPX calcula- tions,i.e.mean absorbed doses to ROIs or absorbed doses to each voxel, and the total number of histories to simulate. The MCNPX input file is finally generated from the geometry, the source distribution and these last parameters.

2.4 Voxel absorbed doses

At the end of the Monte Carlo simulations, the MCNPX output file is processed using OEDIPE to get either mean ab- sorbed doses to ROIs or absorbed doses on the voxel scale.

By default, these results are related to the cumulated activity over an infinite period of time for the specified activity in the field of view. If absorbed doses on the voxel scale have been extracted, a 3D viewer interface, presented in Figure3, is gen- erated to display isodose curves superimposed on the voxel phantom along with mean absorbed doses to ROIs. Further- more, control tools enable one to choose the number and val- ues of isodose curves and to adjust their smoothing, filling or transparency level.

2.5 Voxel biologically effective doses (BEDs)

On the 3D viewer, the user can select the radiobiological module. For the time being, BEDs on the voxel scale can be calculated in the case of a protracted irradiation with a decay- ing source in the absence of washout; the source decay is thus only related to the isotope radioactive decay. When switching to this module, the user will be allowed to set, for each ROI, tissue-specific radiobiological parameters, such as theα/βratio and the sub-lethal damage repair period (Tp). The 3D distribu- tion of BEDs can then be calculated from the absorbed dose

Fig. 4.Histogram viewer interface with optimization tools for the display of DVHs and BVHs.

distribution using either equation 1 or 2 (Dale,1985) depend- ing on whether the user wants to define a specific duration of irradiation T or to consider an infinite treatment time,i.e.as long as the activity is present in the body. If the duration T is at least ten-fold the radioactive decay half-life, equation (1) can be approximated with equation (2). In these equations, D is the voxel absorbed dose in Gy,λis the radioactive decay constant in s⁻¹,µis the repair constant in s⁻¹ defined asµ= ln(2)/T_p,αis the radiosensitivity per unit dose in Gy⁻¹, βis the radiosensitivity per unit square dose in Gy⁻²and T is the treatment time in s.

BEDT =D×

⎡⎢⎢⎢⎢⎢

⎢⎢⎣1+2Dλ² μ−λ

β α

×

2λ1

1−e^−2λT

−_μ+λ¹

1−e^−T(μ+λ) 1−e^−λT

⎤⎥⎥⎥⎥⎥

⎥⎥⎦ (1) BED_∞=D×

1+ Dλ μ+λ

β α

. (2)

The 3D viewer (Fig.3) is then updated to display isoBED curves superimposed on the voxel phantom along with mean BEDs to ROIs.

2.6 Dose- and BED-Volume Histograms

Cumulative dose-volume histograms (DVHs) or BED- volume histograms (BVHs) can be obtained depending on whether absorbed doses or BEDs are displayed on the 3D viewer. These histograms, along with mean, minimum and

maximum absorbed doses or BEDs to ROIs, are displayed on the histogram viewer interface, presented in Figure4. The user can select the ROIs for which plots will be displayed and whether absolute or relative quantities should be plotted. Plot formats can be modified by the user before being saved as pic- tures using standard file formats. A text file containing the mean, minimum and maximum values of absorbed doses or BEDs to ROIs can also be exported.

2.7 Equivalent Uniform Biologically Effective Dose (EUD)

The EUD can be calculated from the BVH viewer inter- face. The EUD is calculated using equation (3) (O’Donoghue, 1999), whereαis the radiosensitivity per unit dose in Gy⁻¹, Ψis the BED distribution and P(Ψ) is the probability density function of the BED.α-values for each ROI are set by the user and differential BVHs, calculated from the cumulative BVHs, are normalized to get the probability density function P(Ψ).

The EUD is finally calculated for each ROI.

EU D=−1 αln

_∞

0

P(Ψ)e^−αΨdΨ

. (3)

2.8 Treatment planning optimization tools

OEDIPE can be used for treatment planning optimization using tools available from the histogram viewer (Fig. 4) for both the absorbed dose (D) and BED modules. Specific tools

280 A. Petitguillaumeet al.: Radioprotection 49(4), 275-281 (2014) enable one to calculate either the volume fraction of a ROI

receiving a given dose (D or BED) or the dose received by a given ROI fraction. Optimization tools can be used to calculate the MIA that meets a tolerance criterion for a ROI. For both modules, the tolerance criterion can be defined on the mean dose, on the volume fraction receiving a given dose or on the maximal dose; an example is presented in Figure4for the cal- culation of the MIA related to a mean absorbed dose to healthy liver which should not exceed 30 Gy. Tolerance criteria de- fined on the volume fraction will be of interest for organs with a “parallel” architecture, such as the liver or lungs, whereas those defined on the maximal dose will be of interest for organs with a “serial” structure, such as the spinal cord (SFRO,2007).

For the radiobiological module, as the BED is not proportional to the injected activity, a dichotomy algorithm is used which stops when the tolerance criterion is met with an error infe- rior to the accepted error keyed in by the user. Furthermore, to study fractionation protocols, an additional option was devel- oped to calculate the MIA for a tolerance criterion and a frac- tionation protocol defined as a number of fractions and the dis- tribution of the total activity among these fractions. The user can then either choose a protocol with time delays between fractions of at least ten-fold the radioactive decay half-life or define specific time delays. In the second case, the BED dis- tribution is calculated using equation (1) for each fraction and considering both the residual activity from the previous frac- tions and the additional activity of the new fraction. However, for the irradiation delivered after the last administered fraction, the BED distribution is calculated using equation (2). Once the MIA has been calculated, DVHs and BVHs can be obtained and displayed, along with mean, minimum and maximum ab- sorbed doses and BEDs, for this optimal activity and the spec- ified fractionation protocol. Finally, EUDs can also be calcu- lated; the total EUD for fractionated protocols being equal to the sum of EUDs delivered by each individual fraction.

3 Results and discussion

In nuclear medicine, OEDIPE has been used for both diag- nostic and therapeutic applications. Up to now, the main pur- pose of the studies conducted was either dosimetry or treat- ment planning optimization. In terms of dosimetry, absorbed doses delivered by commonly used diagnostic radiopharma- ceuticals for the latest ICRP reference computational phan- toms (Hadid et al., 2013) were calculated using OEDIPE.

A personalized Monte Carlo dosimetry was then performed for knee treatments with⁹⁰Y-synovectomy (O’Dohertyet al., 2014). The results were used to evaluate the performance of other methods, such as OLINDA and dose-kernel techniques, for these treatments. Furthermore, OEDIPE was used to study the potential of personalized Monte Carlo dosimetry for treat- ment planning optimization in selective internal radiation ther- apies (SIRT) (Petitguillaumeet al.,2014). In this specific con- text, predictive dosimetry was performed from ^99mTc-MAA evaluations and used to calculate the MIA with different tol- erance criteria on OARs. In particular, the availability of a tridimensional distribution of absorbed doses allowed consid- ering tolerance criteria defined on DVHs; these criteria being

of interest to take advantage of the parallel characteristics of the OARs,i.e.the lungs and nontumoral liver in SIRT. Con- sidering the computational time, the automation of several steps in OEDIPE and the time delay between the^99mTc-MAA evaluation and the ⁹⁰Y-microsphere treatment, the discussed methodology would be compatible with the specific treatment workflow and thus applicable in clinical routine for treatment planning in SIRT. Furthermore, the potential added value of personalized Monte Carlo dosimetry integrating the evaluation step and treatment planning optimization in targeted radionu- clide therapy was confirmed.

The OEDIPE radiobiological module will be of interest to establish BED-effect relationships with pertinent values of ra- diobiological parameters from clinical outcomes in targeted ra- dionuclide therapies. Moreover, both treatment planning op- timization and treatment evaluation in targeted radionuclide therapies will be achievable using OEDIPE. In terms of treat- ment planning optimization, the availability of the BED dis- tribution and the associated EUD will enable one to calculate the MIA according to BED-based tolerance criteria. In such a way, it will also be possible to compare potential biologi- cal effects resulting from different therapeutic options, such as EBRT, brachytherapy or other fractionation protocols.

4 Conclusion

The OEDIPE software is a user-friendly interface with an automation level suitable for personalized Monte Carlo dosimetry for nuclear medicine applications in clinics. For tar- geted radionuclide therapies, this advanced dosimetry can be used for both treatment planning and treatment efficacy eval- uation. Specific tools are available to define the anatomy and the activity distribution for direct Monte Carlo calculations of the absorbed dose distribution. A new module has been im- plemented to calculate the biologically effective dose (BED) distribution and the equivalent uniform biologically effective doses (EUDs) to ROIs. Optimization tools are available, as a help for treatment planning optimization, to calculate the maximal injectable activity that could be administered to the patient while still meeting tolerance criteria on organs at risk (OARs). These tools can be used to design fractionation pro- tocols and evaluate them in terms of the mean BED and EUD to OARs and tumoral lesions. Further improvements are on- going to automatically model the patient-specific biokinetics from serial emission data.

References

Bolch W.E., Bouchet L.G., Robertson J.S., Wessels B.W., Siegel J.A., Howel R.W., Erdi A.K., Aydogan B., Costes S., Watson E.E.

(1999) MIRD Pamphlet No.17: The dosimetry of nonuniform activity distributions – Radionuclide S values at the voxel level, J. Nucl. Med.40, 11S-36S.

Bolch W.E., Eckerman K.F., Sgouros G., Thomas S.R. (2009) MIRD Pamphlet No.21: A generalized schema for radiopharmaceutical dosimetry – Standardization of nomenclature,J. Nucl. Med.50, 477-484.

CEU - Council of the European Union (2013) Council direc- tive 2013/59/EURATOM laying down basic safety standards for protection against the dangers arising from exposure to ionising radiation, and repealing Directives 89/618/Euratom, 90/641/Euratom, 96/29/Euratom, 97/43/Euratom and 2003/ 122/Euratom, 5 December 2013, p. 25.

Chiavassa S., Aubineau-Lanièce I., Bitar A., Lisbona A., Barbet J., Franck D., Jourdain J.R., Bardiès M. (2006) Validation of a per- sonalized dosimetric evaluation tool (OEDIPE) for targeted ra- diotherapy based on the Monte Carlo MCNPX code, Phys. Med.

Biol.51, 601-616.

Cremonesi M., Ferrari M., Bartolomei M., Orsi F., Bonomo G., Arico D., Mallia A., De Cicco C., Pedroli G., Paganelli G. (2008) Radioembolisation with⁹⁰Y-microspheres: dosimetric and radio- biological investigation for multi-cycle treatment,Eur. J. Nucl.

Med. Mol. Imaging35, 2088-2096.

Dale R.G. (1985) The application of the linear-quadratic dose- effect equation to fractionated and protracted radiotherapy,Br. J.

Radiol.58, 515-528.

Dale R.G. (1989) Radiobiological assessment of permanent implants using tumour repopulation factors in the linear-quadratic model, Br. J. Radiol.62, 241-244.

De Carlan L., Aubineau-Lanièce I., Lemosquet A., Borissov N., Jourdain J.R., Jeanbourquin D., Le Guen B., Franck D. (2003) Application of new imaging and calculation techniques to activity and dose assessment in the case of a¹⁰⁶Ru contaminated wound, Radiat. Prot. Dosim.105, 219-223.

EANM (2013)Radionuclide Metabolic Therapy - Clinical Aspects, Dosimetry and Imaging, ISBN: 978-3-902785-08-4.

Franck D., Laval L., Borissov N., Guillierme P., Bordy J.M.

(2001) Development of voxelised numerical phantoms using MCNP Monte Carlo code: Application to in vivo measurement, Radioprotection36, 77-86.

Glatting G., Bardiès M., Lassman M. (2013) Treatment planning in molecular radiotherapy,Z. Med. Phys.23, 262-269.

Hadid L., Gardumi A., Desbrée A. (2013) Evaluation of absorbed and effective doses to patients from radiopharmaceuticals using the ICRP 110 reference computational phantoms and ICRP 103 formulation,Radiat. Prot. Dosim.156(2), 141-159.

ICRP Publication 38 (1993) Radionuclide transformations – Energy and intensity of emissions,Ann. ICRP11-13.

O’Doherty J., Clauss R., Scuffham J., Khan A., Petitguillaume A., Desbrée A. (2014) Three dosimetry models of Lipoma Arborescens treated by ⁹⁰Y synovectomy, Med. Phys. 41 (5), 052501.

O’Donoghue J.A. (1999) Implications of nonuniform tumor doses for radioimmunotherapy,J. Nucl. Med.40, 1337-1341.

Petitguillaume A., Bernardini M., Hadid L., de Labriolle-Vaylet C., Franck D., Desbrée A. (2014) Three-dimensional personalized Monte Carlo dosimetry in ⁹⁰Y-microspheres therapy of hep- atic metastases: non tumoral liver and lungs radiation protec- tion considerations and treatment optimization,J. Nucl. Med.55, 405-413.

RCR - The Royal College of Radiologists (2006) Fractionation in ra- diotherapy: A brief history. In:Radiotherapy dose-fractionation (Board of Faculty of Clinical Oncology, book auth.) pp. 10-13.

SFRO (2007) Guide des procédures de radiothérapie externe, http://www.sfro.org/sfro_pro/media/pdf/guide_procedure_

radiotherapie_2007.pdf.

Sgouros G., Kolbert K.S., Sheikh A., Pentlow K.S., Mun E.F., Barth A., Robbins R.J., Larson S.M. (2004) Patient-specific dosimetry for¹³¹I thyroid cancer therapy using ¹²⁴I PET and 3-dimensional-internal dosimetry (3D-ID) software, J. Nucl.

Med.45, 1366-1372.

Wu Q., Mohan R., Niemierko A., Schmidt-Ullrich R. (2002) Optimization of intensity-modulated radiotherapy plans based on the equivalent uniform dose,Int. J. Radiat. Oncol. Biol. Phys.52, 224-235.

Cite this article as:A. Petitguillaume, M. Bernardini, D. Broggio, C. de Labriolle Vaylet, D. Franck, A. Desbrée. OEDIPE, a software for personalized Monte Carlo dosimetry and treatment planning optimization in nuclear medicine: absorbed dose and biologically effective dose considerations. Radioprotection 49(4), 275-281 (2014).